same game again and again

[john916zhang]

Hi there,

Thanks for developing this wonderful tools. But I have a question: when I let Mugo match against itself, the game of each time is exactly same. How to let Mugos play differently?

Thanks again.

[brilee]

Try using 'randompolicy' as a option instead of 'policy'. Policy will take the most likely move, but randompolicy will select randomly, weighted by the NN's output probabilities.